What a website is actually for.

Ask most owners what their website is for and the answer is some version of “so people can find us”. That is a real job. It is also the one with the most competition, the slowest payback and the least control over the result. Meanwhile the unglamorous work — knowing what is in stock, telling a client where their order is, collecting the same five fields by email forty times a week — runs on spreadsheets, phone calls and an inbox. A website is the cheapest place a business already owns to put that work. The numbers say the gap is real: in 2025, 79.01% of EU enterprises had a website, but only 53.47% used any ERP, CRM or business-intelligence software at all, and among small enterprises ERP use falls to 41.08% against 88.71% for large ones. Nearly everyone has the front door. Far fewer have anything behind it.

What tends to pay back first

A rough ordering, not a rule — the right first build is whichever job currently costs the most hours.

The brochure assumption

A brochure site answers one question — who are you and how do we contact you — and then does nothing until someone fills in a form. That is a fine thing to own. The problem is that it quietly sets the ceiling on what anyone expects from the site, so every other job stays manual by default.

The useful reframe is to stop asking what the website should say, and start asking which repetitive thing in the business it could absorb. Almost every answer below started life as somebody’s spreadsheet.

Selling, including when there is no checkout

The obvious use is selling directly, and for a retail business it needs no argument. The less obvious version is B2B, where a “buy” button is wrong but the process is still mechanical: a configurator that produces a real quote, a reorder page for existing accounts, a price list that respects the customer’s negotiated tier.

Both replace an email thread with a transaction. The only difference is whether money changes hands at the end of it.

Knowing what you actually have: stock and inventory

This is the job most often left on a spreadsheet, and the one with the sharpest payback. A site connected to real stock data can show live availability instead of prompting a phone call, warn you before a line runs out, refuse to sell something you cannot ship, and keep quantities honest across a shop, a marketplace and a physical counter that all sell from the same shelf.

None of that requires a full ERP. It requires one source of truth that everything else reads from — which is precisely what an integration layer is for. Eurostat puts ERP use at 41.08% among small enterprises against 88.71% among large ones: the capability is not exotic, it is just unevenly distributed by company size.

Orders, documents and giving clients their own login

Every “where is my order?” email is a job the site could have done. A client area — order status, invoices, delivery notes, past quotes, the files you keep re-sending — turns a support conversation into a page somebody reads at 11pm without involving you.

It also changes how the relationship feels. We run exactly this for our own clients as a client portal, because the alternative is a folder in an inbox that only one person can find.

Booking, scheduling and the calendar problem

For anyone selling time rather than objects — clinics, studios, trades, consultants — the website replaces the back-and-forth of finding a slot. Availability, deposits, reminders and cancellations all become self-service, and no-show rates usually improve because the reminder is automatic rather than remembered.

The value here is not the booking form. It is that the calendar stops living in one person’s head.

Internal tools: the spreadsheet nobody fully trusts

Most businesses run on a handful of shared spreadsheets that have quietly become critical: a job board, a stock sheet, a pricing matrix, a status tracker. They work until two people edit at once, or until the person who wrote the formulas leaves.

A small internal tool behind a login — with real permissions, an audit trail, and validation that refuses impossible input — often costs a fraction of a public-facing redesign, and it is the part of the system the team actually touches every day. This is ordinary web development; it just never appears in the brochure.

Work that should not need a person

Once information arrives through the site rather than an inbox, it can be routed without anyone retyping it: an enquiry that lands in the CRM already tagged, a quote that generates itself from the configurator, a follow-up that goes out on day three whether or not anyone remembered.

That is what automation means in practice — not replacing judgement, but removing the copying-between-systems that eats a morning. The test for whether something qualifies is simple: if a person is transcribing data from one screen into another, that step is a candidate.

Answering the same question for the hundredth time

Support volume is rarely varied. A documented answer, a search that works, and a well-scoped assistant handle the recurring 80% and leave the genuinely unusual cases to a human — which is the only part where a human adds anything.

The measurable version of this is tickets that never get opened, so record the baseline before you build. Without it you will have no way to prove it worked.

Hiring, and being checked out

A careers page is a filter as much as an advert: candidates who read how you work self-select before anyone spends an interview slot. The same is true of buyers. In B2B, a serious purchase is preceded by someone quietly checking whether you look like a real company that will still exist next year — and they do it without ever contacting you.

That audience never fills in a form, so it is invisible in your lead numbers. It still decides deals.

The audience and the data you own

Followers are rented. Reach on any platform is adjusted by someone else’s algorithm and pricing, and a channel that worked last year can quietly stop working. An email list, a customer account, your own analytics and your own search logs are assets that cannot be repriced without your consent.

The search terms people type into your own site are the most underused of these — an unfiltered list of what customers want and cannot find, written by them, for free.

Where to start

Not all of it, and probably not in the order above. Pick the job that currently costs the most hours or loses the most orders, and build only that. The table below is the rough shape of what tends to pay back first, by business type. If the site itself is the bottleneck rather than any one feature, Website Upgrade is usually a cheaper route than a rebuild.

Sources

The enterprise adoption figures quoted above are Eurostat’s, from the 2025 survey on ICT usage and e-commerce in enterprises. Checked August 2026.

• Eurostat — Digital economy and society statistics: enterprises ↗

  In 2025, 79.01% of EU enterprises had a website — 95.65% of large enterprises against 76.66% of small ones.

• Eurostat — E-business integration ↗

  ERP, CRM and BI adoption for 2025: 53.47% of EU enterprises used any of the three, CRM specifically 28.51%, and ERP 41.08% among small enterprises against 88.71% among large ones.

• Eurostat — Share of enterprises using any business software (ERP, CRM, BI) ↗

  The data table behind those shares, filterable by country and enterprise size — useful for checking Romania or Germany specifically rather than the EU average.